Skip to contentSkip to conversation

block: don't forget user settings

PATCH Applied Keith Busch 6 participants 11 messages Jan 5, 2023 – Jan 9, 2023 Block layer

lore
Share & tools
lore
#0 lore raw

v4 cover block: don't forget user settings

From: Keith Busch <[email protected]>

If the user overrides the max sectors for their device (which is currently defaulting to quite a low value for modern hardware), their request is lost on a rescan. Save it and fix the weird type issues.

Changes since v3: Fixed the unsigned long/unsigned int issue raised by clang.

Keith Busch (2):
block: make BLK_DEF_MAX_SECTORS unsigned
block: save user max_sectors limit

 Documentation/ABI/stable/sysfs-block |  3 ++-
 block/blk-settings.c                 |  9 +++++++--
 block/blk-sysfs.c                    | 21 +++++++++++++++------
 drivers/block/null_blk/main.c        |  3 +--
 include/linux/blkdev.h               |  4 +++-
 5 files changed, 28 insertions(+), 12 deletions(-)
Signature · 1 line
2.30.2
#1 lore raw

PATCH v4 1/2 block: make BLK_DEF_MAX_SECTORS unsigned

From: Keith Busch <[email protected]>

This is used as an unsigned value, so define it that way to avoid having to cast it.

---
 block/blk-settings.c          | 2 +-
 drivers/block/null_blk/main.c | 3 +--
 include/linux/blkdev.h        | 3 ++-
 3 files changed, 4 insertions(+), 4 deletions(-)
block/blk-settings.c +1 -1
index 0477c4d527fee..9875ca131eb0c 100644
@@ -135,7 +135,7 @@ void blk_queue_max_hw_sectors(struct request_queue *q, unsigned int max_hw_secto
 	limits->max_hw_sectors = max_hw_sectors;
 
 	max_sectors = min_not_zero(max_hw_sectors, limits->max_dev_sectors);
-	max_sectors = min_t(unsigned int, max_sectors, BLK_DEF_MAX_SECTORS);
+	max_sectors = min(max_sectors, BLK_DEF_MAX_SECTORS);
 	max_sectors = round_down(max_sectors,
 				 limits->logical_block_size >> SECTOR_SHIFT);
 	limits->max_sectors = max_sectors;
drivers/block/null_blk/main.c +1 -2
index 7d28e3aa406c2..4c601ca9552a0 100644
@@ -2123,8 +2123,7 @@ static int null_add_dev(struct nullb_device *dev)
 	blk_queue_physical_block_size(nullb->q, dev->blocksize);
 	if (!dev->max_sectors)
 		dev->max_sectors = queue_max_hw_sectors(nullb->q);
-	dev->max_sectors = min_t(unsigned int, dev->max_sectors,
-				 BLK_DEF_MAX_SECTORS);
+	dev->max_sectors = min(dev->max_sectors, BLK_DEF_MAX_SECTORS);
 	blk_queue_max_hw_sectors(nullb->q, dev->max_sectors);
 
 	if (dev->virt_boundary)
include/linux/blkdev.h +2 -1
index 43d4e073b1115..2b85161e22561 100644
@@ -1095,11 +1095,12 @@ static inline bool bdev_is_partition(struct block_device *bdev)
 enum blk_default_limits {
 	BLK_MAX_SEGMENTS	= 128,
 	BLK_SAFE_MAX_SECTORS	= 255,
-	BLK_DEF_MAX_SECTORS	= 2560,
 	BLK_MAX_SEGMENT_SIZE	= 65536,
 	BLK_SEG_BOUNDARY_MASK	= 0xFFFFFFFFUL,
 };
 
+#define BLK_DEF_MAX_SECTORS 2560u
+
 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
 {
 	return q->limits.seg_boundary_mask;
Signature · 1 line
2.30.2
#5 lore raw

PATCH v4 2/2 block: save user max_sectors limit

In reply to Keith Busch

From: Keith Busch <[email protected]>

The user can set the max_sectors limit to any valid value via sysfs /sys/block/<dev>/queue/max_sectors_kb attribute. If the device limits are ever rescanned, though, the limit reverts back to the potentially artificially low BLK_DEF_MAX_SECTORS value.

Preserve the user's setting as the max_sectors limit as long as it's valid. The user can reset back to defaults by writing 0 to the sysfs file.

---
 Documentation/ABI/stable/sysfs-block |  3 ++-
 block/blk-settings.c                 |  9 +++++++--
 block/blk-sysfs.c                    | 21 +++++++++++++++------
 include/linux/blkdev.h               |  1 +
 4 files changed, 25 insertions(+), 9 deletions(-)
Documentation/ABI/stable/sysfs-block +2 -1
index cd14ecb3c9a5a..ac1e519272aa2 100644
@@ -432,7 +432,8 @@ Contact:	[email protected]
 Description:
 		[RW] This is the maximum number of kilobytes that the block
 		layer will allow for a filesystem request. Must be smaller than
-		or equal to the maximum size allowed by the hardware.
+		or equal to the maximum size allowed by the hardware. Write 0
+		to use default kernel settings.
 
 
 What:		/sys/block/<disk>/queue/max_segment_size
block/blk-settings.c +7 -2
index 9875ca131eb0c..9c9713c9269cc 100644
@@ -40,7 +40,7 @@ void blk_set_default_limits(struct queue_limits *lim)
 	lim->virt_boundary_mask = 0;
 	lim->max_segment_size = BLK_MAX_SEGMENT_SIZE;
 	lim->max_sectors = lim->max_hw_sectors = BLK_SAFE_MAX_SECTORS;
-	lim->max_dev_sectors = 0;
+	lim->max_user_sectors = lim->max_dev_sectors = 0;
 	lim->chunk_sectors = 0;
 	lim->max_write_zeroes_sectors = 0;
 	lim->max_zone_append_sectors = 0;
@@ -135,7 +135,12 @@ void blk_queue_max_hw_sectors(struct request_queue *q, unsigned int max_hw_secto
 	limits->max_hw_sectors = max_hw_sectors;
 
 	max_sectors = min_not_zero(max_hw_sectors, limits->max_dev_sectors);
-	max_sectors = min(max_sectors, BLK_DEF_MAX_SECTORS);
+
+	if (limits->max_user_sectors)
+		max_sectors = min(max_sectors, limits->max_user_sectors);
+	else
+		max_sectors = min(max_sectors, BLK_DEF_MAX_SECTORS);
+
 	max_sectors = round_down(max_sectors,
 				 limits->logical_block_size >> SECTOR_SHIFT);
 	limits->max_sectors = max_sectors;
block/blk-sysfs.c +15 -6
index 93d9e9c9a6ea8..5486b6c57f6b8 100644
@@ -239,19 +239,28 @@ static ssize_t queue_zone_append_max_show(struct request_queue *q, char *page)
 static ssize_t
 queue_max_sectors_store(struct request_queue *q, const char *page, size_t count)
 {
-	unsigned long max_sectors_kb,
+	unsigned long var;
+	unsigned int max_sectors_kb,
 		max_hw_sectors_kb = queue_max_hw_sectors(q) >> 1,
 			page_kb = 1 << (PAGE_SHIFT - 10);
-	ssize_t ret = queue_var_store(&max_sectors_kb, page, count);
+	ssize_t ret = queue_var_store(&var, page, count);
 
 	if (ret < 0)
 		return ret;
 
-	max_hw_sectors_kb = min_not_zero(max_hw_sectors_kb, (unsigned long)
+	max_sectors_kb = (unsigned int)var;
+	max_hw_sectors_kb = min_not_zero(max_hw_sectors_kb,
 					 q->limits.max_dev_sectors >> 1);
-
-	if (max_sectors_kb > max_hw_sectors_kb || max_sectors_kb < page_kb)
-		return -EINVAL;
+	if (max_sectors_kb == 0) {
+		q->limits.max_user_sectors = 0;
+		max_sectors_kb = min(max_hw_sectors_kb,
+				     BLK_DEF_MAX_SECTORS >> 1);
+	} else {
+		if (max_sectors_kb > max_hw_sectors_kb ||
+		    max_sectors_kb < page_kb)
+			return -EINVAL;
+		q->limits.max_user_sectors = max_sectors_kb << 1;
+	}
 
 	spin_lock_irq(&q->queue_lock);
 	q->limits.max_sectors = max_sectors_kb << 1;
include/linux/blkdev.h +1 -0
index 2b85161e22561..b87ed829ab941 100644
@@ -288,6 +288,7 @@ struct queue_limits {
 	unsigned int		max_dev_sectors;
 	unsigned int		chunk_sectors;
 	unsigned int		max_sectors;
+	unsigned int		max_user_sectors;
 	unsigned int		max_segment_size;
 	unsigned int		physical_block_size;
 	unsigned int		logical_block_size;
Signature · 1 line
2.30.2
#6 lore raw

Re: [PATCHv4 2/2] block: save user max_sectors limit

Quoted 11 lines · Keith Busch
block/blk-sysfs.c +2 -1
index 93d9e9c9a6ea8..5486b6c57f6b8 100644
@@ -239,19 +239,28 @@ static ssize_t queue_zone_append_max_show(struct request_queue *q, char *page)
  static ssize_t
  queue_max_sectors_store(struct request_queue *q, const char *page, size_t count)
  {
-	unsigned long max_sectors_kb,
+	unsigned long var;
+	unsigned int max_sectors_kb,

block/blk-sysfs.c static ssize_t queue_zone_append_max_show(struct request_queue *q, char *page) · line 251

Quoted 10 lines
block/blk-sysfs.c +2 -2
  		max_hw_sectors_kb = queue_max_hw_sectors(q) >> 1,
  			page_kb = 1 << (PAGE_SHIFT - 10);
-	ssize_t ret = queue_var_store(&max_sectors_kb, page, count);
+	ssize_t ret = queue_var_store(&var, page, count);
 
  	if (ret < 0)
  		return ret;
 
-	max_hw_sectors_kb = min_not_zero(max_hw_sectors_kb, (unsigned long)
+	max_sectors_kb = (unsigned int)var;

Shouldn't this function report an error if 'var' is too large to fit into an unsigned int?

Otherwise this patch looks good to me.

Thanks,

Bart.

#7 lore raw

Re: [PATCHv4 2/2] block: save user max_sectors limit

Quoted 15 lines · Bart Van Assche
Quoted 14 lines · Keith Busch
block/blk-sysfs.c +2 -2
index 93d9e9c9a6ea8..5486b6c57f6b8 100644
@@ -239,19 +239,28 @@ static ssize_t queue_zone_append_max_show(struct request_queue *q, char *page)
  static ssize_t
  queue_max_sectors_store(struct request_queue *q, const char *page, size_t count)
  {
-	unsigned long max_sectors_kb,
+	unsigned long var;
+	unsigned int max_sectors_kb,
  		max_hw_sectors_kb = queue_max_hw_sectors(q) >> 1,
  			page_kb = 1 << (PAGE_SHIFT - 10);
-	ssize_t ret = queue_var_store(&max_sectors_kb, page, count);
Quoted 8 lines

block/blk-sysfs.c static ssize_t queue_zone_append_max_show(struct request_queue *q, char *page) · line 251

Quoted 5 lines
block/blk-sysfs.c +2 -1
+	ssize_t ret = queue_var_store(&var, page, count);
  	if (ret < 0)
  		return ret;
-	max_hw_sectors_kb = min_not_zero(max_hw_sectors_kb, (unsigned long)
+	max_sectors_kb = (unsigned int)var;

Shouldn't this function report an error if 'var' is too large to fit into an unsigned int?

Yes it should, and queue_var_store() will return -EINVAL if that happens. That's certainly not clear just from reading this patch, but the condition is handled.

Reviewed-by on v4 2/2
#9 lore raw

Re: [PATCHv4 2/2] block: save user max_sectors limit

In reply to Keith Busch

Keith,

Quoted 8 lines

The user can set the max_sectors limit to any valid value via sysfs /sys/block/<dev>/queue/max_sectors_kb attribute. If the device limits are ever rescanned, though, the limit reverts back to the potentially artificially low BLK_DEF_MAX_SECTORS value.

Preserve the user's setting as the max_sectors limit as long as it's valid. The user can reset back to defaults by writing 0 to the sysfs file.

Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
maintainer Applied
#10 lore raw
In reply to Keith Busch
Quoted 8 lines · Keith Busch

If the user overrides the max sectors for their device (which is currently defaulting to quite a low value for modern hardware), their request is lost on a rescan. Save it and fix the weird type issues.

Changes since v3: Fixed the unsigned long/unsigned int issue raised by clang.

[...]

Applied, thanks!

[1/2] block: make BLK_DEF_MAX_SECTORS unsigned

      commit: c749b9c6de40f0882d9c83c8a3780e631603eb9d
[2/2] block: save user max_sectors limit
      commit: 39001023bf1dfcee0ad1602501fbdea6f0d3d3bb

Best regards,

Signature · 1 line
Jens Axboe